Cardiac Left Ventricular Myocardial Tissue Density, Evaluated by Computed Tomography and Autopsy Alexandra G

Gheorghe et al. BMC Medical Imaging (2019) 19:29 https://doi.org/10.1186/s12880-019-0326-4

RESEARCHARTICLE Open Access Cardiac left ventricular myocardial tissue density, evaluated by computed tomography and autopsy Alexandra G. Gheorghe1* , Andreas Fuchs2, Christina Jacobsen1, Klaus F. Kofoed2,3, Rasmus Møgelvang2, Niels Lynnerup1 and Jytte Banner1

Abstract Background: Left ventricular mass (LVM) is an independent risk factor for the prediction of cardiac events. Its assessment is a clinically important diagnostic procedure in cardiology and may be performed by Computed Tomography (CT). The aim of this study was to assess the correlation between the cardiac left ventricular shell volume (LVShV) determined by postmortem Computed Tomography (PMCT) and the anatomic LVM obtained at autopsy and to calculate the myocardial tissue density. Methods: A total of 109 deceased individuals were examined with a 64-slice CT scanner and LVShV was determined. At autopsy, the left ventricle was dissected and weighted. The correlation between LVShV and the anatomic LVM was analysed. Asymmetric left ventricular (LV) hypertrophy was recorded. Inter-observer variability was evaluated, and a density value for myocardial tissue was calculated. Results: The mean age of the deceased was 55 ± 16 years, and 58% was men. We found 30 cases of asymmetric LV hypertrophy. A highly positive correlation existed between LVShV and anatomic LVM (r = 0.857; p < 0.0001), regardless of hypertrophy, asymmetric hypertrophy and gender. The mean difference in the inter-observer variability for LVShV assessment was - 4.4 ml (95% CI: -26.4; 17.6). A linear regression analysis was performed, resulting in a value of 1.265 g/ml for myocardial tissue density. Applying the hitherto used myocardial tissue density of 1.055 g/ml underestimated the anatomic LVM by 18.1% (p < 0.0001). Conclusion: PMCT is a helpful tool for the assessment of LVM, and LVShV is highly correlated with LVM as assessed by subsequent autopsy. The correlation between the two was independent of gender, hypertrophy and LV asymmetric hypertrophy. We found a higher myocardial tissue density of 1.265 g/ml compared to previous studies. We show that PMCT combined with autopsy may contribute not only to anatomical but also clinical knowledge. Keywords: Left Ventricular Shell Volume, Left Ventricular Mass, Anatomic Left Ventricular Mass, Computed Tomography, Left Ventricular Mass, Myocardial Tissue Density, Post Mortem Computed Tomography

Background specific region of the LV, asymmetric hypertrophy, re- Left ventricular mass (LVM) is a prognostic factor in gardless of the overall heart size [6, 7]. The most wide- cardiac disease [1], and thus, assessment of LVM is used spread tool for non-invasive measurements of the LVM in the diagnosis and risk stratification of patients in clin- is echocardiography [4, 7]. LVM determination by differ- ical practice [2]. In addition, abnormal patterns of left ent imaging modalities is based on the LV shell volume ventricular (LV) size and shape have been found to have (LVShV), which is the difference between the epicardial prognostic relevance [3–5]. Hypertrophy may occur in a and endocardial volumes [4, 8]. The LVShV is subse- quently converted to mass by multiplying it with the density of myocardial tissue [8, 9]. The clinically accepted * Correspondence: [email protected] 1Section of Forensic Pathology, Department of Forensic Medicine, University value of myocardial tissue density is 1.055 g/ml [9–14]; of Copenhagen, Frederik V’s vej 11, 1 sal, 2100 Copenhagen, Denmark however, there are different published values [6, 8, 15–22], Full list of author information is available at the end of the article

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Gheorghe et al. BMC Medical Imaging (2019) 19:29 Page 2 of 9

cf. Table 1. LVM has been calculated based on Computed Post mortem CT imaging protocol Tomography (CT)-angiography [10], magnetic resonance Non-contrast PMCT imaging was performed using a imaging (MRI) [23] and echocardiography [24], but the myo- 64-slice Multi Detector Computed Tomography scanner cardial density has not been validated. The aim of this study (MDCT) (Siemens Somatom definition syngo 2010A; Sie- was to assess the association between the LVShV determined mens medical solutions, Forchheim, Germany). The follow- by postmortem Computed Tomography (PMCT) and the ing scanning parameters were used: tube current time: 300 anatomic LVM obtained at autopsy, and to calculate the mAs, tube voltage: 120 kV, slice collimation: 32 × 0.6 mm myocardial tissue density. and slice acquisition: 64 × 0.6 mm. The iterative recon- structed slice thickness was 1.5 mm using a soft tissue con- Methods volution kernel. All images were acquired with the deceased Study population and design wrapped in an artefact-free body bag in a supine position The study population is part of a prospective nationwide with elevated arms (except for one due to a BMI of 50). The autopsy-based Danish forensic study, SURVIVE-let the bodies were placed in the CT gantry and scanned from head dead help the living, which focuses on deceased with to toe. After reconstruction of the raw data, the images were mental illness [25]. The deceased were autopsied at the transferred to the local PACS server. Section of Forensic Pathology at the Institute of Forensic Medicine, University of Copenhagen over an Imaging analysis eleven-month period, from January 2014 to November Assessment of the myocardial tissue was performed with 2014. The study group comprised 116 deceased that fulfilled commercial software (Vitrea 6.3, Vital Images, Inc., MN, the following inclusion criteria: individuals with a known or USA). Imaging analysis was performed by individuals who a suspected mental illness, and the intake of antipsychotic, were blinded to the autopsy results. The non-contrast antidepressive or anti-anxiety medication or a suspicion PMCT data were reviewed in multiplanar reconstructions thereof. The exclusion criteria were putrefaction, suspicion after adjusting the planes to a long-axis view of the heart. ofacriminalact,inadequateCTimagequalityfortheana- TheLVmyocardialtissuewassegmentedbymanuallytra- lysis and failure to scan due to severe obesity. This led to the cing the endocardial and epicardial border for each long-axis exclusion of seven individuals. After a medical inquest, the de- slice and adding up the corresponding volumes from each ceased underwent a whole-body PMCT. The scans occurred slice (Fig. 1). We used normal cardiac window settings with within 24 h prior to autopsy. At the autopsy (see details a level = 200 Hounsfield Units (HUs) and width = 1000. Vox- below), the sex, age, weight and height of the deceased were els corresponding to myocardial tissue were identified using recorded, as were the size, shape and macroscopic changes of mean threshold attenuation values of 50 ± 10 HUs to distin- the heart. The project was approved by the Danish National guish it from the LV blood pool (Fig. 1). The region of inter- Committee on Health Research Ethics (1377517). est (ROI) for the segmentation of both LV myocardial tissue

Table 1 Previous studies investigating the myocardial tissue density or correlation between an image modality and the LVM Studies describing the myocardial tissue density or myocardial tissue gravity Reference Year Material Method Number included Density/gravity Correlation, r Friedmann C [21] 1951 Cadaver human hearts Submersion in water 45 1.029 Masshoff W [14] 1967 Cadaver human hearts Submersion in water 23 1.055 Webb A [35] 1979 Equine hearts Archimedes principle 18 1.033 Schapira JN [36] 1981 Canine hearts Phased array sector scan 15 1.04 Rufeng B [20] 2007 Cadaver human hearts Density Instrument 169 1.3, 0.9 Studies referring to a myocardial tissue density factor based on previous studies Snyder [18] 1975 Human hearts Radiology – 1.033 Wyatt A [31] 1979 Canine hearts Echocardiography 21 1.05 0.94, 0.92 Reicheck [17] 1981 Cadaver human hearts Echocardiography, ekg 34 1.04 Schiller [29] 1983 Canine Echocardiography 10 1.055 0.98 Deveraux [12] 1986 Cadaver human hearts Echocardiography 55 1.04 0.92 Manning J [30] 1990 Rat hearts MRI 28 1.055 0.98 Jackowski C [13] 2004 Cadaver MRI, CT-angio 80 1.05 Lang [8] 2005 Living Echocardiography 85 1.04 Fuchs A [10] 2016 Living CT-angiography 569 1.055 Gheorghe et al. BMC Medical Imaging (2019) 19:29 Page 3 of 9

Fig. 1 Long-axis wiev of the heart, illustrating the myocardial tissue (1) and the LV blood pool (2). The orange- and blue line indicates the epicardial and endocardial contours, respectively and the LV blood pool was defined at the height of the papillary muscles were preserved, and the hearts valves septum and the LV outlet by HUs. The ROI was set at a were removed. The epicardial fat was resected. The LV minimum of 1 cm2. Depending on the heart size, the num- slices were blotted dry and separately weighed using an ber of manually traced slices varied from 70 to 100 in each electronic scale (Mettler Toledo, ICS425, Glostrup, case. Finally, based on the segmentation, the LVShV was cal- Denmark) with a 1-g precision. culated in millilitres and converted to mass, by multiplying it The heart was recorded as being hypertrophic according with 1.055 g/ml. Inter-observer variability was assessed in 25 to international forensic pathology standards and charts [28]. randomly chosen individuals. Fifteen randomly chosen LV Anterior, lateral, posterior and septal wall thicknesses were myocardial tissues were re-analysed using a different cardiac measured at the level of the midventricular transversal slice. setting with level = 45 HU and width = 50 HU. Papillary muscles were excluded from this measurement. Asymmetric LV hypertrophy was defined as biggest heart LV autopsy procedure wall (millimetres) > 1.3*thinnest heart wall (millimetres) [7]. The autopsies were performed in accordance with the extended autopsy protocol of the SURVIVE study and Statistical analyses the departments guidelines and were accredited by the Data analysis was performed using the SPSS statistical soft- Danish accreditation fund (DANAK). The length of the ware package (IBM Corp. IBM SPSS Statistics for Windows, corpses were measured in the supine position from the Version 24, USA). Data was presented as the mean ± stand- top of the head (vertex) to the heel using an inelastic ard deviation, and a p-value of < 0.05 was considered statisti- measuring tape. The heart was removed during the aut- cally significant. Data distribution was tested with the opsy by preserving an adequate extension of the base Ancova/ linear regression test. Bland-Altman analysis was and pulmonary vessels and dissected according to inter- used to assess the re-analysis of LVShV and the degree of national standards detailed by Basso et al. [26]. After the inter-observer variability. Scatterplots were used to illustrate removal of clotted blood, the whole heart was weighed. the level of agreement and linearity between the CT obtained Dissection of the left ventricle was performed according LVShV and the anatomic LVM as well as asymmetric LV to the method described by Bove et al. [27] with modifi- hypertrophy. The degree of correlation was determined by cations; an incision was made in the posterior part of Pearson’s correlation coefficient. The association between the the left atrium following the atrioventricular groove to CT-obtained LVShV and the anatomic LVM, with and with- remove the atria from the ventricles. The ventricles were out adjustment for gender, was assessed by linear regression cut into transverse slices (short-axis direction). The right analysis. A density factor was calculated by linear regression ventricle was separated from the septum and the left analyses, forcing the slope through origo (as done in compar- ventricle. The trabeculae encrusted in the septum and able studies) [7, 8]. Gheorghe et al. BMC Medical Imaging (2019) 19:29 Page 4 of 9

Table 2 Study group data, autopsy measurements and LVShV obtained by CT Women Men Mean 1 SD Range Mean 1 SD Range Age 59 18 21–94 52 14 22–79 Body weight, kg 67 67.4 38–130 79 15 54–120 Anatomic LVM, g 163 35 96–277 217 50 140–326 Anterior wall thickness, mm 13 3 9–18 13 3 8–20 Lateral wall thickness, mm 13 4 8–20 14 3 9–20 Posterior wall thickness, mm 13 4 9–25 13 3 9–18 Septal thickness, mm 14 3 9–20 14 4 8–21 Total heart weight, g 354 77 231–582 459 104 302–812 LVShV, ml 126.4 30.7 77–218 168.9 43.5 101–282 Descriptive data for study group and anatomic measures and CT determined LVShV (LVM, left ventricular mass; g, grams; mm, millimeters; kg, kilograms; LVShV, LV Shell volume)

Results Left ventricular mass and LVShV Data were collected from a total study population of 109 The results for anatomic LVM, LV wall thickness and deceased, comprising 46 females (42%) and 63 males LVShV are given in Table 2. There were 36 hypertrophic (58%). The age and body weight distributions are de- hearts, (25 men and 11 women) based on total heart tailed in Table 2. The deceased were autopsied at a mean weight. Thirty cases had asymmetric LV hypertrophy (19 time (± 1 SD) of 44 ± 16 h (range 29–150 h) after the men and 11 women) of which thirteen had hypertrophic declaration of death. hearts (total weight).

Fig. 2 Correlation between the CT determined LVShV and anatomic LVM (Pearson r = 0.874, p < 0.0001) stratified by the presence of asymmetric LV hypertrophy (triangles) Gheorghe et al. BMC Medical Imaging (2019) 19:29 Page 5 of 9

Table 3 Correlation between the CT determined LVShV and performed linear regression analyses with LVShV as the ex- anatomic LVM planatory variable and LVM as the dependent variable. n R value P value Stratification by gender showed no differences. Potentially, Women 46 0.739 <0.0001 the resultant regression equation (genders combined) could Men 63 0.836 <0.0001 thus be used for calculating LVM when LVShV has been determined. Traditionally, clinical volume-to-mass calculations All 109 0.857 <0.0001 use only a density factor, i.e. the volume is simply multiplied Asymmetric LV hypertrophy 30 0.874 <0.0001 by a density factor [8, 9]. This is in line with a theoretical ap- Correlation values by gender and asymmetric LV hypertrophy proach to a density factor: zero volume equals zero mass. Earl- ier attempts at establishing a myocardial density factor have Overall, the CT-determined LVShV was highly correlated therefore, more or less explicitly, assumed that a volume-to- with the anatomic LVM (Pearson r = 0.857, p < 0.0001), also mass regression must go through origo [7, 12, 24], hence allowing us to assume linearity. Significant correlation was resulting in a simple factor, and not in a regression equation also found when the cohort was stratified by the presence of with both a factor and a constant. We therefore performed asymmetric LV hypertrophy (Fig. 2,Table3) and by gender new linear regression analyses, but this time forcing the line of (Table 3). The inter-observer mean difference in LVShV as- regression through the origo. Again, differences by gender sessment by CT was - 4.4 ml (95% CI: -26.4; 17.6). The mean were negligible, thus allowing us to propose the resultant slope HUs value of the ROI was 47 ± 7 in the segmentation of for the combined genders, 1.265 g/ml, as the myocardial tissue myocardial tissue and 68 ± 13 in the LV blood pool. A linear density. This density value is based on CT-determined LVShV regression equation was calculated for each gender (Table 4) and the actual weight of the anatomic LVM obtained at aut- and for the genders combined (Fig. 3,Table4). Linear regres- opsy. When we compared this new myocardial tissue density sion analyses were also performed by forcing the line of re- value with the hitherto used value in the clinical setting of gression through the origo for each gender (Table 4)andfor 1.055 g/ml (6–15), we found that the latter value significantly the genders combined (Fig. 3,Table4). The differences be- underestimated the anatomic LVM. tween the models based on gender were negligible. Forcing Indeed, the assumed density of myocardial tissue has var- the line of regression through the origo resulted in the fol- ied over time, e.g., 1.029, 1.03, 1.04 and 1.055 g/ml, not least lowing estimated model: LVM = LVShV*1.265 g/ml, with because of differing techniques in determining the density, 1.265 g/ml as the myocardial tissue density. The R2 values e.g., by immersing cardiac muscle tissue or hearts in water and residual standard error are presented in Table 4.Hyper- (Archimedes principle) [14, 20, 21], different image modal- trophic hearts based on total heart weight did not alter the ities and different animal species [12, 17, 29–32](cf.Table1). myocardial tissue density compared to non-hypertrophic Echocardiography is the most widespread tool used for hearts (data not shown). Using the hitherto clinically ac- the quantification of LVM [7, 8], CT- angiography is often cepted myocardial tissue density of 1.055 g/ml, the anatomic used for LVM calculations [10] and cardiac MRI is often LVM was underestimated by 18.1% (95% CI = 29.9–40.0 g), considered the gold standard for LVM assessment [23, 33]. p < 0.0001, (Fig. 3,Table5).ThemeandifferenceinLVShV Although not often used, non-contrast CT can also be used assessment by CT with the different settings in 15 randomly for information in LV size in the clinical setting [34]. In this chosen LV myocardial tissues was - 19.4 ml (95% CI: -65.3; large study of recently deceased individuals, we showed that 25.5), p >0.5. the use of non-contrast PMCT for the determination of LVShV is a useful tool and has a satisfactory repeatability Discussion due to the fact that a clear distinction between the HUs of As expected, our data showed a highly positive correlation myocardial tissue and LV blood pool without active circula- between the CT-determined LVShV and the anatomic LVM; tion could be made. However, these CT settings were opti- also when stratifying by gender and when focusing on cases mized for contrast-based studies. Changing the CT settings with hypertrophy and asymmetric LV hypertrophy. We then in 15 randomly chosen cases showed an overall, but

Table 4 Linear regression for the LVM by LVShV Linear Regression Equation R2 Residual SE Linear regression forced through origo Equation R2 Residual SE Women 45.73 + 0.93057 g/ml * LVShV 0.9835 26.15 1.267 g/ml* LVShV 0.9792 29.25 Men 59.63 + 0.93057 g/ml * LVShV 0.9835 26.15 1.264 g/ml* LVShV 0.9792 29.25 All 42.36 + 1.0061 g/ml * LVShV 0.7331 26.73 1.265 g/ml* LVShV 0.9792 29.11 There was no significant difference between the slope of the genders, why a fitted model was used, 0.93057. A slightly different constant was seen due to the LVM being bigger in men. There was no difference in the LVM calculating equation (LVShV * myocardial density), when the line of regression was forced through the origo. Knowing the gender and not forcing the linear regression through the origo had the best fit. (LVM, left ventricular mass; g, grams; ml, milliliters; LVShV, LV shell volume) Gheorghe et al. BMC Medical Imaging (2019) 19:29 Page 6 of 9

Fig. 3 Regression equations lines for LVM by LVShV. The lines are for genders combined (with the constant of 42.36). CT_vol represents the CT determined LVShV and the autopsy mass represents the anatomic LVM. The plot also shows the underestimation of the anatomic LVM when applying the clinically used myocardial tissue density of 1.055g/ml negligible and non-significant (p >0.5),decreaseinLVShV LVShV and LVM obtained at autopsy. This may explain why (mean difference − 19.4ml).Wedidfindthattheimage our value differs from other proposed values. Since our value quality improved in some cases using these different set- is higher, this means that a given LVShV will result in a tings (n = 7/15), as a more clear-cut distinction between the higher LVM, and will be most pronounced for higher and blood-pool and the myocardium was achieved. However, in potentially pathological LVShV, which may be of interest for other cases a lot of noise was introduced, so we find that clinicians and warrants further studies. the majority of the scans were in fact better analysed with Although it is currently not fully accepted clinically to the original settings. use the LVM as a regular test for patients [3], the exact CT Several studies have presented normal reference values measurements and accurate values for myocardial tissue for LVM based on CT-angiography [10, 11, 15], echocar- density can lead to important therapeutic opportunities diography [8, 24] and MRI [23, 33] and thus it is reason- concerning diagnosis, treatment and prognostics. If the able to develop modality-specific reference values. hitherto used density value is substituted with our value, However, regardless of the imaging modality used to ob- this could have implications for the reference values as it tain the LVShV, all shell volumes are converted to mass will move some patients to higher LVMs indicating abnor- by multiplying it with the myocardial tissue density [8], mal LVM. If one continues with the hitherto used density with 1.055 g/ml being the most commonly used density. value, there will be no problem in using the reference inter- To our knowledge, this study is the first to calculate the hu- vals, but to the best of our knowledge, this would mean that man myocardial density factor using non-contrast CT based the recorded LVM is not the real LVM.

Table 5 LVM calculated using different myocardial tissue density and the anatomic LVM Women Men All Mean 1 SD Range Mean 1 SD Range Mean 1SD Range LVM, g (LVShV*1.055 g/ml) 133 32 81–230 178 46 107–298 159 46 81–298 LVM, g (LVShV*1.265 g/ml) 160 39 97–276 213 55 127–357 191 55 97–357 Anatomic LVM, g 163 35 96–277 217 50 140–326 194 51 96–326 Resultant LVM using the different equations, with comparison to the anatomic LVM (LVM, left ventricular mass; g, grams; ml, milliliters; LVShV, LV shell volume) Gheorghe et al. BMC Medical Imaging (2019) 19:29 Page 7 of 9

Study limitations Finally, our calculation of a new value for myocardial The following study limitations need to be addressed. This density is made performing a linear regression on the study included 73 non-hypertrophic hearts, but in order to LVM and LVShV values of our study population. Thus, develop LVM reference values based on the density, a bigger theoretically, we cannot be sure that very small hearts or study population would be preferable. We did not take into LV masses below 96 g will conform to the linear model. consideration the status of ischaemic heart disease, fat infil- Thiscanprobablyonlybeinvestigatedbyalsoapplying tration or LV fibrosis and how this potentially could affect our method to a subadult study population. However, we the size and shape of the left ventricle. The papillary muscles note that all other analyses on myocardial tissue density were included in the LVShV measurements. However, the were also constrained or even more limited. We would impact of the inclusion or exclusion of papillary muscles on also note that when comparing with the hitherto-used the assessment of LV function is negligible [15]. The scans value for myocardial density the differences are most were non-contrast scans, which in a clinical setting makes pronounced for bigger hearts, so even if our new value the differentiation between the LV blood pool and the LV has not been tested on very small hearts, any differences myocardium difficult. However, non-contrast scans are pos- might be assumed minor. sible in deceased individuals (cf. Fig. 1). We chose to perform the present study without contrast to avoid extravasation of Conclusion contrast media in the examined deceased individuals and The unique access to both PMCT scans and autopsy mea- thereby omit a possible weight effect on the myocardium. surements allowed us to assess the correlation between AstudybyBaiRetal.[20]hassuggestedatendencyof LVShVandtheanatomicLVM;theanalysiswasalsopossible lower myocardial tissue density associated with pathological based on gender and LV asymmetric hypertrophy. Our re- changes as oedema, none of our included cases had under- gression models determined only very small differences went putrefaction, as these were excluded. Therefore, we do when stratified by gender and negligible differences when assume that post mortal changes did not have any impact on forcing the regression through the origo, which allowed us the myocardial tissue density calculation. In vivo, the myo- to determine a myocardial density at 1.265 g/ml. Applying cardium consist of intra-myocardial blood-volume. Postmor- the hitherto used myocardial tissue density value (1.055 g/ tem, this volume may change, e.g., postmortem extravasation ml) significantly underestimated the LVM. Our proposed or, conversely, fluid accumulation from leaking and decom- new value is the result of post-mortem CT for volume deter- posing endocardial structures. Such changes are small and mination, followed by post-mortem dissection for obtaining usually only become pronounced with extended postmortem LV weight. We argue that this allows for a more precise de- intervals [35]. The deceased individuals in this study were termination of these two basic parameters, but obviously, we kept at reduced temperature and autopsied rather quickly are aware that post-mortem anatomy may not be directly after declaration of death. Morphological observations as translational to clinical studies. Several LVM reference value well as quantitative results suggest that elements of the blood tables have been produced using myocardial tissue density are resistant to autolytic effects [36]. Overall, this leads us to value of 1.055 g/ml, thus continuing to use this value when assume that no significant organ volumetric changes took converting from LVShV to LVM will not have immediate place. Water displacement was not performed in this study. clinical implications. However, we do think that our study There are several factors to take into consideration when calls for critical evaluation of especially high LVMs and how performing water displacement measurements/techniques this value is obtained. on cadaver hearts. Although it works accurately with solid objects, biological tissue is by its nature permeable and may Abbreviations be compressed or distended, and it may be fixed or unfixed. CT: Computed Tomography; DANAK: Danish Accreditation fund; HUs: Hounsfield Units; LV: Left Ventricular; LVM: Left Ventricular Mass; Given that we wanted to investigate CT-derived volume LVShV: Left ventricular shell volume; MDCT: Multi Detector Computer measures,asthisiswhatisusedclinically,wehencechoseto Tomography; MRI: Magnetic Resonance Imaging; PMCT: Post Mortem base our volumes on this method. Boundaries are sensitive Computed Tomography; ROI: Region of Interest to threshold and windowing. We used a HU threshold for Acknowledgments myocardial tissue of 50 ± 10, and 1 mm slice thickness for The authors thank the physicians, morgue technicians, laboratory assistants CT evaluations. The CT settings used for this study are opti- and medical students for their assistance in the SURVIVE project as well as mized for contrast-based studies, and a different cardiac win- statistician Kyle Raymond for his statistical knowledge and assistance. dow settings could result in different myocardial volume. Funding Also, different thresholds may apply for epicardial and endo- This project was sponsored by the Department of Forensic Medicine, cardial borders and partial volume effects may be important. University of Copenhagen (Unit number 361100). Due to the relatively close HU values of the blood pool and Availability of data and materials the myocardium, the myocardium may have been overesti- The datasets used and/or analyzed during the current study are available mated in some cases and underestimated in other cases. from the corresponding author on reasonable request. Gheorghe et al. BMC Medical Imaging (2019) 19:29 Page 8 of 9

Authors’ contributions general population study. Eur Heart J Cardiovasc Imaging. 2016;17:1009–17. AGG, AF, CJ KFK, NL and JB contributed to the design of the study. AGG https://doi.org/10.1093/ehjci/jev337. contributed to the literature search and the image analysis. AF contributed 11. Hindsø L, Fuchs A, Kühl JT, Nilsson EJP, Sigvardsen PE, Køber L, et al. Normal to literature search, the inter-observer imaging analysis and as Vitrea advisor. values of regional left ventricular myocardial thickness, mass and AGG and CJ contributed to the CT-scan protocol. AF, KFK and RM contrib- distribution-assessed by 320-detector computed tomography angiography uted as cardiology advisors. JB and CJ contributed as forensic pathology ad- in the Copenhagen general population study. Int J Card Imaging. 2017;33: visors. NL contributed as statistical advisor. AGG conducted the LVM excision 421–9. https://doi.org/10.1007/s10554-016-1015-9. at autopsy. AGG contributed to the manuscript writing and all drafts. All au- 12. Devereux RB, Alonso DR, Lutas EM, Gottlieb GJ, Campo E, Sachs I, et al. thors read and approved the final manuscript. Echocardiographic assessment of left ventricular hypertrophy: comparison to necropsy findings. Am J Cardiol. 1986;57:450–8. Ethics approval and consent to participate 13. Jackowski C, Schweitzer W, Thali M, Yen K, Aghayev E, Sonnenschein M, et al. The next of kin gave informed written or verbal consents that was approved Virtopsy: postmortem imaging of the human heart in situ using MSCT and MRI. by the National Committee on Danish Health Research Ethics who also Forensic Sci Int. 2005;149:11–23. https://doi.org/10.1016/j.forsciint.2004.05.019. approved (20-01-2014) the research protocol (version 2.0 number 1377517). 14. Masshoff W, Scheidt D, Reimers HF. Quantitative Bestimmung des Fett-und Myokardgewebes im Leichenherzen. Virchows Arch. 1967;342:184–9. Consent for publication 15. Stolzmann P, Scheffel H, Leschka S, Schertler T, Frauenfelder T, Kaufmann Not applicable. PA, et al. Reference values for quantitative left ventricular and left atrial measurements in cardiac computed tomography. Eur Radiol. 2008;18:1625– Competing interests 34. https://doi.org/10.1007/s00330-008-0939-4. The authors declare that they have no competing interests. 16. Lakatta EG, Mitchell JH, Pomerance A, Rowe GG. Human aging: changes in structure and function. J Am Coll Cardiol. 1987;10:42A–7A. Publisher’sNote 17. Reichek N, Devereux RB. Left ventricular hypertrophy: relationship of anatomic, echocardiographic and electrocardiographic findings. Circulation. Springer Nature remains neutral with regard to jurisdictional claims in 1981;63:1391–8. https://doi.org/10.1161/01.CIR.63.6.1391. published maps and institutional affiliations. 18. Snyder WS, Cook MJ, Nasset ES, Karhausen LR, Howells GP. Tipton IH. Report of the Task Group on Reference Man. n.d.:112–7. Author details 1Section of Forensic Pathology, Department of Forensic Medicine, University 19. Bardeen CR. Determination of the size of the heart by means of the x-rays. – of Copenhagen, Frederik V’s vej 11, 1 sal, 2100 Copenhagen, Denmark. Dev Dyn. 1918;23:423 87. 2Department of Cardiology, The Heart Centre, Rigshospitalet, University of 20. Bai R, Yu X, Wang D, Lv J, Xu G, Lai X. The densities of visceral organs and – Copenhagen, Copenhagen, Denmark. 3Department of Radiology, Diagnostic the extent of pathologic changes. Am J Forensic Med Pathol. 2009;30:148 Centre, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark. 51. https://doi.org/10.1097/PAF.0b013e3181875ab2. 21. Friedman CE. Heart volume, myocardial volume and total capacity of the Received: 8 February 2018 Accepted: 12 March 2019 heart cavities in certain chronic heart diseases; a clinic, roentgenologic and patho-anatomic investigation of the problem of cardiac hypertrophy and dilatation and amount of residual blood of the heart. Acta Medica Scand – References Suppl. 1951;257:1 100. 1. Hoang K, Zhao Y, Gardin JM, Carnethon M, Mukamal K, Yanez D, et al. LV 22. de la Grandmaison GL, Clairand I, Durigon M. Organ weight in 684 adult autopsies: – mass as a predictor of CVD events in older adults with and without new tables for a Caucasoid population. Forensic Sci Int. 2001;119:149 54. metabolic syndrome and diabetes. JACC Cardiovasc Imaging. 2015;8:1007– 23. Gandy SJ, Waugh SA, Nicholas RS, Simpson HJ, Milne W, Houston 15. https://doi.org/10.1016/j.jcmg.2015.04.019. JG. Comparison of the reproducibility of quantitative cardiac left 2. Armstrong AC, Gidding S, Gjesdal O, Wu C, Bluemke DA, Lima JAC. LV mass ventricular assessments in healthy volunteers using different MRI assessed by echocardiography and CMR, cardiovascular outcomes, and scanners: a multicenter simulation. J Magn Reson Imaging. 2008;28: – medical practice. JACC Cardiovasc Imaging. 2012;5:837–48. https://doi.org/ 359 65. https://doi.org/10.1002/jmri.21401. 10.1016/j.jcmg.2012.06.003. 24. Takeuchi M, Nishikage T, Mor-Avi V, Sugeng L, Weinert L, Nakai H, et al. 3. Litwin SE. Cardiac “Morphomics.”. JACC Cardiovasc Imaging. 2015;8:1016–8. Measurement of left ventricular mass by real-time three-dimensional https://doi.org/10.1016/j.jcmg.2015.06.008. echocardiography: validation against magnetic resonance and comparison 4. Celebi AS, Yalcin H, Yalcin F. Current cardiac imaging techniques for with two-dimensional and m-mode measurements. J Am Soc Echocardiogr. – detection of left ventricular mass. Cardiovasc Ultrasound. 2010;8:19. 2008;21:1001 5. https://doi.org/10.1016/j.echo.2008.07.008. 5. Udelson JE. Left ventricular shape. JACC: Heart Failure. 2017;5:179–81. 25. Banner J. SURVIVE: let the dead help the living—an autopsy-based cohort https://doi.org/10.1016/j.jchf.2017.01.005. study for mapping risk markers of death among those with severe mental 6. Emanuel R, Marcomichelakis J, Withers R, O’Brien K. Asymmetric septal illnesses. Scandinavian journal of forensic science n.d.; 24/2018:7–17. https:// hypertrophy and hypertrophic cardiomyopathy. Br Heart J. 1983;49:309. doi.org/10.2478/sjfs-2018-0002. 7. Losi M-A, Nistri S, Galderisi M, Betocchi S, Cecchi F, Olivotto I, et al. 26. Basso C, Burke M, Fornes P, Gallagher PJ, De Gouveia RH, Sheppard M, et al. Echocardiography in patients with hypertrophic cardiomyopathy: usefulness Guidelines for autopsy investigation of sudden cardiac death. Pathologica. of old and new techniques in the diagnosis and pathophysiological 2010;102:391–404. assessment. Cardiovasc Ultrasound. 2010;8:7. 27. Bove KE, Rowlands DT, Scott RC. Observations on the assessment of cardiac 8. Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pellikka PA, et al. hypertrophy utilizing a chamber partition technique. Circulation. 1966;33:558–68. Recommendations for chamber quantification: a report from the American 28. Lucena J, Garcia-Pavia P, Pulpon LA. Clinico-pathological atlas of Society of Echocardiography’s guidelines and standards committee and the cardiovascular diseases. vol. 2015. Springer International Publishing, chamber quantification writing group, developed in conjunction with the Switzerland. 362 pages. page 210-215. ISBN 13:978-3319111452. European Association of Echocardiography, a branch of the European 29. Schiller NB, Skiôldebrand CG, Schiller EJ, Mavroudis CC, Silverman NH, Society of Cardiology. J Am Soc Echocardiogr. 2005;18:1440–63. https://doi. Rahimtoola SH, et al. Canine left ventricular mass estimation by two- org/10.1016/j.echo.2005.10.005. dimensional echocardiography. Circulation. 1983;68:210–6. 9. Fuchs A, Kühl JT, Lønborg J, Engstrøm T, Vejlstrup N, Køber L, et al. 30. Manning WJ, Wei JY, Fossel ET, Burstein D. Measurement of left ventricular Automated assessment of heart chamber volumes and function in patients mass in rats using electrocardiogram-gated magnetic resonance imaging. with previous myocardial infarction using multidetector computed Am J Phys Heart Circ Phys. 1990;258:H1181–6. tomography. Journal of Cardiovascular Computed Tomography. 2012;6:325– 31. Wyatt HL, Heng MK, Meerbaum S, Hestenes JD, Cobo JM, Davidson RM, et al. 34. https://doi.org/10.1016/j.jcct.2012.01.006. Cross-sectional echocardiography. I. Analysis of mathematic models for 10. Fuchs A, Mejdahl MR, Kühl JT, Stisen ZR, Nilsson EJP, Køber LV, et al. Normal quantifying mass of the left ventricle in dogs. Circulation. 1979;60:1104–13. values of left ventricular mass and cardiac chamber volumes assessed by 32. Ward SR, Lieber RL. Density and hydration of fresh and fixed human skeletal muscle. 320-detector computed tomography angiography in the Copenhagen J Biomech. 2005;38:2317–20. https://doi.org/10.1016/j.jbiomech.2004.10.001. Gheorghe et al. BMC Medical Imaging (2019) 19:29 Page 9 of 9

33. Juergens KU, Fischbach R. Left ventricular function studied with MDCT. Eur Radiol. 2006;16:342–57. https://doi.org/10.1007/s00330-005-2888-5. 34. Dykun I, Mahabadi AA, Lehmann N, Bauer M, Moebus S, Jöckel K-H, et al. Left ventricle size quantification using non-contrast-enhanced cardiac computed tomography--association with cardiovascular risk factors and coronary artery calcium score in the general population: the Heinz Nixdorf recall study. Acta Radiol. 2015;56:933–42. https://doi.org/10.1177/0284185114542996. 35. Zilg B, Thelander G, Giebe B, Druid H. Postmortem blood sampling- comparison of drug concentrations at different sample sites. Forensic Sci Int. 2017;278:296–303. https://doi.org/10.1016/j.forsciint.2017.07.006. 36. Penttilä A, Laiho K. Autolytic changes in blood cells of human cadavers. II. Morphological studies. Forensic Sci Int. 1981;17:121–32.